AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

ALM-MTA: Front-Door Causal Multi-Touch Attribution Method for Creator-Ecosystem Optimization

Yuguang Liu, Luyao Xia, Hu Liu, yanzhangxi, Jian Liang, Han Li, Kun Gai

ICLR 2026 · Probabilistic and scientific modeling › Causal discovery · approach: Constraint-based structure learning with latents

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace direct proxy observation with adversarially learned mediator (ablation)ALM-MTA · Production logs offlineAUC: 0.97 -> 0.86 (UAUC 0.90 -> 0.71)Mixed
add contrastive overlap learner (ablation)ALM-MTA with adversarial mediator · Production logs offlineUAUC: 0.71 -> 0.825 (AUC 0.86 -> 0.907)Improved
front-door ALM-MTA vs backdoor DML baselineattribution model · Production logs offlineAUC: 0.6498 -> 0.9070Improved
enrich proxy Y' from strict element match to full path set (ablation)ALM-MTA · Production logs grouped AUUCgrouped AUUC: 0.8496 -> 0.8686Improved
ALM-MTA vs causalMTA on CriteoALM-MTA · Criteolog-loss: AUC 0.9659 -> 0.9729; log-loss 0.0517 -> 0.0634Mixed
ALM-MTA vs causalMTA under tight budgetALM-MTA · Criteo budget 1/16CPA: 25.47 -> 22.03Improved
ALM-MTA vs causalMTA CVR at half budgetALM-MTA · Criteo budget 1/2CVR: 0.1247 -> 0.1248No clear effect
deploy ALM-MTA labels onlineproduction ranking · Online A/BDAU: +0.04%Improved
ALM-MTA vs deployed attribution methodonline attribution · Online attribution accuracyprecision: 4.32% -> 21.88%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy